usda-vision/SESSION_SUMMARY.md

Session Summary: MQTT Auto-Recording Debugging & Theme Implementation

Overview

This session focused on debugging MQTT-triggered auto-recording functionality and implementing dark/light theme toggle in the management dashboard. Multiple issues were identified and resolved related to camera recording, MQTT message handling, and UI theming.

---

Major Tasks Completed

1. Scheduling Module Extraction (Previous Session)

• Extracted the Scheduling module into a separate microfrontend (scheduling-remote)
• Configured Module Federation for the new microfrontend
• Updated docker-compose.yml to include the new service

2. Dark/Light Theme Toggle Implementation

• Created useTheme hook (management-dashboard-web-app/src/hooks/useTheme.ts)
• Added theme toggle button to TopNavbar.tsx
• Applied dark mode classes across multiple components:
  • DashboardLayout.tsx - Main content area
  • Login.tsx - Login page background
  • DataEntry.tsx - Widgets, text, borders
  • UserManagement.tsx - Stats cards, tables, forms
  • ExperimentPhases.tsx - Phase cards, badges
  • index.css - Body background
  • index.html - Inline script to prevent FOUC (Flash of Unstyled Content)

3. MQTT Auto-Recording Debugging

Issues Identified:

1. Manual recording buttons not working - "Record" buttons for cameras couldn't start manual recording
2. Automatic recording unavailable - MQTT-triggered auto-recording wasn't functioning
3. Camera2 initialization error - Error code 32774 when initializing camera2
4. Button visibility issues - "Restart Camera" and "Stop Streaming" buttons appeared transparent

Fixes Applied:

A. Transparent Buttons Fix

File: vision-system-remote/src/components/CameraCard.tsx

• Added inline style attributes to "Restart Camera" and "Stop Streaming" buttons
• Explicitly set backgroundColor and color to prevent transparency

B. Camera Initialization Error (32774)

File: camera-management-api/usda_vision_system/camera/monitor.py

• Enhanced error handling for camera initialization error code 32774
• Added logic to check if camera is already in use by existing recorder/streamer
• If error 32774 occurs, camera is marked as "available" with warning instead of failing completely
• Provides better diagnostics for cameras that might be in use by another process

C. StandaloneAutoRecorder Integration

File: camera-management-api/usda_vision_system/recording/standalone_auto_recorder.py

• Problem: StandaloneAutoRecorder was creating its own MQTT client and camera recorders, causing conflicts
• Solution:
  • Modified constructor to accept camera_manager, state_manager, and event_system instances
  • Removed internal MQTT client setup when using shared instances
  • Now subscribes to EventType.MACHINE_STATE_CHANGED events from the event system
  • Uses camera_manager.manual_start_recording() and manual_stop_recording() instead of creating its own recorders
  • Added comprehensive logging for event reception and recording operations

File: camera-management-api/usda_vision_system/main.py

• Updated USDAVisionSystem constructor to pass camera_manager, state_manager, and event_system to StandaloneAutoRecorder

D. Missing Constants Import

File: camera-management-api/usda_vision_system/camera/recorder.py

• Error: NameError: name 'CAMERA_TEST_CAPTURE_TIMEOUT' is not defined
• Fix: Added missing imports:
  • CAMERA_GET_BUFFER_TIMEOUT
  • CAMERA_INIT_TIMEOUT
  • CAMERA_TEST_CAPTURE_TIMEOUT
  • DEFAULT_VIDEO_FPS
  • BRIEF_PAUSE_SLEEP
  • All imported from .constants

E. Recording Routes Enhancement

File: camera-management-api/usda_vision_system/api/routes/recording_routes.py

• Made StartRecordingRequest parameter optional for start_recording endpoint
• Added logic to create default StartRecordingRequest if request is None
• Added INFO-level logging for recording attempts and success/failure
• Improved error messages for manual_start_recording failures

F. MQTT Logging Improvements

Files:

• camera-management-api/usda_vision_system/mqtt/client.py
• camera-management-api/usda_vision_system/mqtt/handlers.py

Changes:

• Changed MQTT message logging from DEBUG to INFO level for visibility in production
• Added detailed logging for:
  • Message reception: 📡 MQTT MESSAGE RECEIVED - Topic: {topic}, Payload: '{payload}'
  • Message processing: 📡 Processing MQTT message for machine '{machine_name}': '{payload}'
  • Payload normalization: 📡 Normalized payload '{payload}' -> '{normalized_payload}'
  • Event publishing: 📡 Publishing MACHINE_STATE_CHANGED event
  • Connection status: 🔗 MQTT CONNECTED with broker details
  • Subscription confirmation: 📋 MQTT subscribed to {count} topics

G. MQTT Test Script

File: camera-management-api/test_mqtt_simple.py (NEW)

• Created standalone MQTT test script to verify connectivity and message reception
• Connects to MQTT broker and subscribes to configured topics
• Displays received messages in real-time with timestamps
• Validates payload format (on/off/true/false/1/0)
• Shows connection statistics

4. Configuration Updates

File: camera-management-api/config.json

• Confirmed log_level is set to INFO
• auto_recording_enabled is set to true

---

Key Files Modified

Frontend (React/TypeScript)

1. management-dashboard-web-app/src/hooks/useTheme.ts - NEW
2. management-dashboard-web-app/src/components/TopNavbar.tsx
3. management-dashboard-web-app/src/components/DashboardLayout.tsx
4. management-dashboard-web-app/src/components/Login.tsx
5. management-dashboard-web-app/src/components/DataEntry.tsx
6. management-dashboard-web-app/src/components/UserManagement.tsx
7. management-dashboard-web-app/src/components/ExperimentPhases.tsx
8. management-dashboard-web-app/src/index.css
9. management-dashboard-web-app/index.html
10. vision-system-remote/src/components/CameraCard.tsx

Backend (Python/FastAPI)

1. camera-management-api/usda_vision_system/recording/standalone_auto_recorder.py
2. camera-management-api/usda_vision_system/main.py
3. camera-management-api/usda_vision_system/camera/recorder.py
4. camera-management-api/usda_vision_system/camera/monitor.py
5. camera-management-api/usda_vision_system/api/routes/recording_routes.py
6. camera-management-api/usda_vision_system/mqtt/client.py
7. camera-management-api/usda_vision_system/mqtt/handlers.py
8. camera-management-api/test_mqtt_simple.py - NEW

---

Current Status

✅ Completed

• Dark/light theme toggle fully implemented
• Transparent buttons fixed
• Camera initialization error handling improved
• StandaloneAutoRecorder properly integrated with shared instances
• Missing constants imported
• Recording routes enhanced with better logging
• MQTT logging improved to INFO level
• MQTT test script created

🔄 Testing Needed

• MQTT Auto-Recording: Needs verification that MQTT messages trigger recording
  • Test script available: camera-management-api/test_mqtt_simple.py
  • Monitor logs for MQTT message flow: docker compose logs api -f | grep -i "mqtt\|📡"
  • Check MQTT status: curl http://localhost:8000/mqtt/status
  • Check recent events: curl http://localhost:8000/mqtt/events?limit=10

📋 Expected MQTT Message Flow

When a machine turns on/off, the following should appear in logs:

1. 📡 MQTT MESSAGE RECEIVED - Message received from broker
2. 📡 Processing MQTT message - Message being processed
3. 📡 Normalized payload - Payload normalization
4. ✅ Published MACHINE_STATE_CHANGED event - Event published to event system
5. 📡 AUTO-RECORDER: Received MACHINE_STATE_CHANGED event - Auto-recorder received event
6. ✅ Started auto-recording - Recording started

---

MQTT Configuration

From config.json:

{
  "mqtt": {
    "broker_host": "192.168.1.110",
    "broker_port": 1883,
    "username": null,
    "password": null,
    "topics": {
      "vibratory_conveyor": "vision/vibratory_conveyor/state",
      "blower_separator": "vision/blower_separator/state"
    }
  }
}

Machine-to-Camera Mapping (from standalone_auto_recorder.py):

• vibratory_conveyor → camera2
• blower_separator → camera1

---

Testing Commands

Test MQTT Connectivity

# Option 1: Standalone test script
docker compose exec api python test_mqtt_simple.py

# Option 2: Check MQTT status via API
curl -s http://localhost:8000/mqtt/status | python -m json.tool

# Option 3: Monitor API logs for MQTT messages
docker compose logs api -f | grep -i "mqtt\|📡"

# Option 4: Check recent MQTT events
curl -s http://localhost:8000/mqtt/events?limit=10 | python -m json.tool

Restart API Container

docker compose restart api

---

Architecture Notes

Event Flow for Auto-Recording

1. MQTT Message Received → MQTTClient._on_message()
2. Message Processed → MQTTMessageHandler.handle_message()
3. State Updated → StateManager.update_machine_state()
4. Event Published → EventSystem.publish(EventType.MACHINE_STATE_CHANGED)
5. Event Received → StandaloneAutoRecorder._on_machine_state_changed()
6. Recording Started/Stopped → CameraManager.manual_start_recording() / manual_stop_recording()

Key Components

• MQTTClient: Connects to MQTT broker, subscribes to topics, receives messages
• MQTTMessageHandler: Processes MQTT messages, normalizes payloads, updates state
• EventSystem: Publishes/subscribes to internal events
• StandaloneAutoRecorder: Subscribes to MACHINE_STATE_CHANGED events, triggers recording
• CameraManager: Manages camera operations including recording

---

Notes for Future Sessions

1. MQTT Testing: If auto-recording still doesn't work, verify:

   • MQTT broker is reachable from API container
   • Topics match exactly (case-sensitive)
   • Messages are being published to correct topics
   • Payload format is recognized (on/off/true/false/1/0)

2. Camera Initialization: Error 32774 typically means camera is in use. The system now handles this gracefully but may need investigation if cameras are not accessible.

3. Theme Persistence: Theme preference is stored in localStorage and persists across sessions.

4. Log Level: Currently set to INFO in config.json for comprehensive event tracking.

---

Session Date

Session completed with focus on MQTT debugging and enhanced logging for troubleshooting auto-recording functionality.